---
title: "Anan's Replication Project - Islamophobia?: Religion, Contact with Muslims, and the Respect for Islam"
author: "Anan Hafez"
date: "5/7/2021"
output:
  pdf_document:
    keep_tex: true
header-includes:
   - \usepackage{dcolumn}
---

```{r setup, include=FALSE}
knitr::opts_chunk$set(echo = TRUE)
library(tidyverse)
library(readxl)
library(stargazer)
library(gtsummary)
library(skimr)
library(gt)
```

# Introduction

&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;Much of the dialogue surrounding the Muslim community in the United States has revolved around the notion of Muslims being violent and threatening to the Western world. In the past two decades, the U.S. military has occupied two Muslim-majority countries, where it faces multiple armed resistance movements, some explicitly Muslim in orientation, others less so. This violence has produced bus and train bombings in Europe, attacks on globally dispersed targets linked to support for U.S. war efforts, and massive new government structures—again, global in scope—dedicated to the detention, questioning, and elimination of suspects and “combatants” who are, with few exceptions, Arab, Muslim, or both. While the characterization of Islam as “terror” was firm long before 9/11, it has grown stronger ever since as high-profile enemies in the war on terror have been defined, and have defined themselves, as Muslim. The result, now recognized by journalists, politicians, intellectuals, and others, is a pervasive “Islamophobia”; which has motivated acts of mosque vandalism, hate crimes against individuals thought to be Muslim, sensational press coverage of “the Muslim threat”, and the selective policing and surveillance of Muslim communities because of the prevalent suspicion that Islam is somehow antithetical to democratic values. 

&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;As a young Muslim-American who grew up in the shadow of 9/11, examining the dynamics of Islamophobia has always been particularly relevant. What contributes the most to the fear of Muslims among Americans, and how can we mitigate it? Fortunately, I am far from the first to contribute to this literature. In a 2012 paper published in the Review of Religious Research titled 'Islamophobia?: Religion, Contact with Muslims, and the Respect for Islam', author Jong Hyun Jung (Purdue University) examines how individual religious backgrounds, beliefs, and personal contact contribute to reduce or Islamophobia.[^1][^2] For example, do larger religious in-groups such as Christian Protestants have lower regard for smaller religious out-groups, in this case, Islam? How does one's view of God affect their esteem towards Muslims? In his paper, Jung draws on a nationally representative survey to explore these hypotheses, the Portraits of American Life Study. The Portraits of American Life Study (PALS) is an in-home survey of the U.S. adult population, consisting of 2,610 respondents with a first wave in 2006 and a follow-up in 2012 (Emerson & Sikkink, 2006). Because the study oversampled some ethnic groups, the survey was weighted to enhance its representativeness of U.S. population as a whole. The 2006, 2012, and merged datasets and summary statisitics are available here: https://www.thearda.com/pals/

&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;Jung lays forward four main hypotheses for how the previously stated individual factors can affect one's view of Muslims or Islam in general. In order, they are:
  
  + **Hypothesis 1:** An individual’s religious affiliation will be related to the degree to which that person respects Islam, with Christians more likely to have low regard for Islam.
  + **Hypothesis 2:** Individuals who hold an image of God who punishes people for their sins will be more likely to have low respect for Islam.
  + **Hypothesis 3:** Contact with Muslims will reduce the odds of having the least respect for Islam.
  + **Hypothesis 4:** Evangelical Protestants are more likely to have the least respect for Islam as they have contact with Muslims more frequently.
  
&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;While Jung briefly lays out his reasoning behind each hypothesis, I find it worthwhile to dive deeper into each theory and lean on the most recent literature to see why each is quite plausible.
  
## Analysis of Hypothesis 1: 

&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;Since the terrorist attacks of September 11th, 2001, public opinion on Muslim-Americans and Islam in general has been negative and unsympathetic. The American Mosaic Project Survey discovered that the vast majority of Americans, about 78%, hold a low regard for Muslims (Edgell, Gerteis, & Hartmann, 2006). A more recent Pew Forum report revealed that 30% of Americans say they have a favorable opinion of Islam while slightly more than 38% have an unfavorable view (Pew Research Center, 2010). A thorough dive into existing literature on an individual’s religious background and their reactions to inter-group contact can help explain why Americans view Islam in this way. 

&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;Religious traditions can often influence the attitudes people hold toward the validity of other groups by encouraging their followers to seek congruence between their belief system and outside world (Wald & Calhoun-Brown, 38, 2011). In this way, religion serves as a strong basis for the formation of group identity in the United States (Wald & Calhoun-Brown, 37, 2011). These religious subcultures often prescribe certain behavioral norms as well as establishing symbolic boundaries for relationships that help one group define themselves against others (Wald & Calhoun-Brown, 40, 2011). As Jung states, "This boundary drawing can strengthen cultural membership that determines insiders and outsiders in terms of legitimacy. Thus, religious differences may form the basis for inclusion and exclusion, affecting how believers in one religious subculture perceive others in different religious traditions". America’s religious history has been defined by the dominance of Christianity, Protestantism in particular, and has shaped the values, institutions, and practices that are common in the U.S. (Marty, 2011). In “Christian America”, non-Christian religious others are considered as not sharing one’s vision of America, thus often labeled as an out-group that fall outside of the Judeo-Christian mainstream of America (Edgell, Gerteis, & Hartmann, 2006). This suggests that people have more favorable evaluations of in-group relative to an out-group, and that out-groups are often the objects of negative opinions and hostile attitudes (Edgell, Gerteis, & Hartmann, 2006). In this case, it is probable that Christian respondents have a lower measure of respect for Islam relative to other groups.

## Analysis of Hypothesis 2:

&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;Individual theological beliefs provide important motivations for how religious people view outside communities. Compassion and loving one’s neighbor are both theological conceptions that can motivate individuals to hold their fellow citizens in higher regard, regardless of their differences. Recent research has illuminated that having an image of God as judgmental or punishing raises the odds that religious adherents hold hostile views towards those outside of their religious group (Wald & Calhoun-Brown, 55, 2011). These findings suggest that adherents who are most willing to engage the external community independent of a place of worship are those with less judgmental images of God. Thus, it not unfair to assume that those with harsher conceptions of God would look less favorably on those they consider to be non-believers, such as Muslims.

## Analysis of Hypothesis 3:

&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;The proportion of Muslims in the United States has nearly doubled since 2007 to 1.1% of the total population encompassing 3.45 million people largely due to immigration and conversion (Pew Research Center, 2018). In the midst of growing level of religious pluralism, Islam has become one of the most visible non-Christian faiths in American society. Research has shown that as members of minority groups grow and become more deeply woven into society, they are more likely to be accepted by general population and to have expanding social capital including interpersonal networks (Putnam & Campbell, 2012). These findings point to more Muslims being engaged in social and interreligious interactions with others. Greater social contact reduces prejudice with that respective group, as Putnam and Campbell argue that “having a religiously diverse social network leads to a more positive assessment of specific religious group” (Putnam & Campbell, 527, 2012). This phenomenon is referred to as “religious bridging” and suggests that prior contact with Muslims can lead to a more welcoming attitude toward Muslims. This mirrors the vast research on interracial contact and its ability to foster acceptance, tolerance, and harmony, particularly between African-Americans and Whites (Campbell, 2006). However, previous research into this hypothesis, specifically as it relates to Muslims, has yielded the opposite result. A study by Khari and Brown, analyzing 2002 telephone interview data from American adults found that contact with Jews but not with Muslims promoted acceptance of religious pluralism among White Christians (Khari & Brown, 2011). It can be inferred that respondents felt that Islam did not fit in with their understanding of religious diversity even if they had come in contact with a Muslim (Khari and Brown, 2011). However, as the telephone data is taken in the year immediately following the September 11th attacks, respondents’ views on Islam and religious diversity may be compromised by the ongoing “War on Terror” outside the U.S. During this period, words such as ‘terrorism’ and ‘terrorist’ in political and media discourses increasingly became subjective terms to disproportionately highlight violence committed by Arabs or Muslims, thus conflating the two (Bakali, 2016). By associating these terms in popular media, distress at the terrorist attacks of 9/11 and the following war seriously impacted public perception of Islam negatively and potentially served as a confounder in the study’s findings. Thus, this justifies continuing with the “religious bridging” hypothesis posed by Putnam and Campbell when examining more recent data of Muslim contact on religious tolerance.

## Analysis of Hypothesis 4:

&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;In some cases, however, it is entirely possible that some religious backgrounds may encourage intolerance for those who hold different beliefs. For some, contact with individuals with seperate religious belief system may reinforce commitment to their own strict belief system and hostility towards that other group. A growing body of research has found these associations in everyday life. Drawing on “racial threat” literature, Campbell finds that for groups whose theology is more conflicted with the outside world, particularly Evangelical Protestants, when confronted with certain religious minorities, will turn towards their theological tradition and strengthen their commitment to their faith, in turn raising hostility towards the out-group (Campbell, 2006). He refers to this phenomenon in his research as an increase in “religious threat”. Just as white voters have been found to feel threatened by the presence of African-Americans in their community and become more likely to vote for racially conservative candidates as the proportion of blacks in their community rises, conservative religious traditions tend to hold groups that don’t share their religious beliefs in lower regard as they come into contact more frequently (Campbell, 2006). For those groups, contact and conflict with non-believers emphasize and strengthen the cohesion of the group and allows their stricter theology to prosper (Finke & Stark, 2005). Although Campbell specifically evaluates the relationship of Evangelicals and secularists, the distinct nature of Evangelical Protestantism and Islam suggests that the “religious threat” hypothesis can be applied in this instance. In fact, research by Cimino on Evangelical rhetoric following the September 11th terrorist attacks on the U.S. found that Evangelical leaders emerged as strong critics and even antagonists of Islam, serving to draw sharper boundaries between Islam and Christianity (Cimino, 2005).

# Data, Methods, and Analysis

&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;Jung only draws from the first wave 2006 PALS results in his paper; I'll look towards expanding his analysis with the 2012 data later in this project. Unfortunately, he also does not provide the code he used to estimate his results or the exact variables he used so I will extrapolate from his explanations and tables.

&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;The dependent variable of interest for Jung's analysis was measured by using responses to Question 540, which asks the following: “Which religion do you have the least respect for?” The choice of “Islam” was specifically offered along with “Buddhism”, “Christianity”, “Hinduism”, “Judaism”, “Mormonism”, “Do not respect any of these”, “Equal respect for all”, and “Other.” As Jung did, I will recode this into a random variable that displays (1) for the choice “Islam” and (0) for all other choices. This is will allow me to look at the proportion of respondents who hold “Islam” in the lowest esteem and fit these responses to a model later on. 

&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;Our independent variables have to do with the respondent’s religious tradition, conception of God, and level of contact they have with Muslims. Question 671 asks for the respondent’s religious tradition, these categories include: Evangelical Protestant, Mainline Protestant, Black Protestant, Catholic, Jewish, Other, Other Protestant, and Unaffiliated. From what I can see, it appears the author recodes each of these responses into its own variable thats is (1) when that affiliation is selected and zero when not. The respondent's conception of God as a punisher is measured using Question 18 which asks,"In the last few years, how often have you felt that God is punishing you for your sins or lack of spirituality?". The response categories here are, "Never", "Rarely", "Sometimes", "Very often", and "Every day or almost every day". Lastly, contact with Muslims is measured in Question 210 which measures how often a respondent has a conversation with a Muslim; ". The response categories are “Never”, “Once or twice a year”, “Once or few times a month”, “Once or few times a week”, and “Everyday”. These response categories will be recoded as a new ordinal random variable with values (1-5) depending on the frequency of the respondent’s contact.

&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;Jung's control variables consist of socio-demographic and religious variables. Many of his socio-demographic variables are dichotomous with gender (1 = Female), marital status (1 = married), race (1 = White), region (1 = South), and political orientation (1 = Republican). Education (Education) and household income (Income) are measured as ordinal variables ranging from less than high school (1) to professional degree (11) and from less than $5,000 (1) to $200,000 or more (19).

&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;Jung also adds an extra control to ensure that the degree to which one respects Islam is not a simple proxy measure for religious fundamentalism. Previous studies have shown that religious fundamentalism is a strong predictor of prejudice towards outsiders. Respondant fundamentalism is measured by Question 26 which asks, "Was your religious text fully inspired by God, partly inspired by God, or not inspired by God?". Jung recodes the answer choices to  fully inspired (1) and all other responses (0).

&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;In the next section, I will clean and mutate the 2006 PALS data to best match Hyon's recoded variables and produce a summary table similar to Table 1 in his paper.

```{r}
# Reading in 2006 PALS Data
PALS_2006 <- read_xlsx("Data/Portraits of American Life Study, 1st Wave, 2006.XLSX")

# Recoding and cleaning data
PALS_2006 <- PALS_2006 %>% select(HR_AGE, RC_2, IC_6, RELTRAD, ARA_5, RACE_ALL, 
                                  HR_GEND, DM_2, PO_6, REGION, LV_1, 
                                  DM_INC, ARBL_3) %>% 
  mutate(Age = HR_AGE) %>% 
  mutate(Muslim_Hostility = if_else(RC_2 == 4, 1, 0)) %>% 
  mutate(Muslim_Contact = case_when(
  IC_6 == 1 ~ 1,
  IC_6 == 2 ~ 2,
  IC_6 == 3 ~ 3,
  IC_6 == 4 ~ 3,
  IC_6 == 5 ~ 4,
  IC_6 == 6 ~ 4,
  IC_6 == 7 ~ 5,
  IC_6 == -7 ~ NA_real_,
  IC_6 == -4 ~ NA_real_)) %>%
  mutate(Evangelical_Protestant = if_else(RELTRAD == 2, 1, 0)) %>% 
  mutate(Mainline_Protestant = if_else(RELTRAD == 3, 1, 0)) %>%
  mutate(Black_Protestant = if_else(RELTRAD == 1, 1, 0)) %>%
  mutate(Catholic = if_else(RELTRAD == 4, 1, 0)) %>%
  mutate(Jewish = if_else(RELTRAD == 5, 1, 0)) %>%
  mutate(Other_Religion = if_else(RELTRAD == 6, 1, 0)) %>%
  mutate(Other_Protestant = if_else(RELTRAD == 8, 1, 0)) %>%
  mutate(Unaffiliated = if_else(RELTRAD == 7, 1, 0)) %>%
  mutate(Punishing_God = case_when(
  ARA_5 == 1 ~ 1,
  ARA_5 == 2 ~ 2,
  ARA_5 == 3 ~ 3,
  ARA_5 == 4 ~ 4,
  ARA_5 == 5 ~ 5,
  ARA_5 == -7 ~ NA_real_,
  ARA_5 == -4 ~ NA_real_)) %>% 
  mutate(White = if_else(RACE_ALL == 1, 1, 0)) %>% 
  mutate(Female = if_else(HR_GEND == 0, 1, 0)) %>% 
  mutate(Education = case_when(
  DM_2 == 1 ~ 1,
  DM_2 == 2 ~ 2,
  DM_2 == 3 ~ 3,
  DM_2 == 4 ~ 4,
  DM_2 == 5 ~ 5,
  DM_2 == 6 ~ 6,
  DM_2 == 7 ~ 7,
  DM_2 == 8 ~ 8,
  DM_2 == 9 ~ 9,
  DM_2 == 10 ~ 10,
  DM_2 == 11 ~ 11,
  DM_2 == -7 ~ NA_real_,
  DM_2 == -4 ~ NA_real_,
  DM_2 == 12 ~ NA_real_)) %>% 
  mutate(Republican = if_else(PO_6 == 2, 1, 0)) %>% 
  mutate(South = if_else(REGION == 3, 1, 0)) %>% 
  mutate(Married = if_else(LV_1 == 1, 1, 0)) %>% 
  mutate(Income = case_when(
  DM_INC == 1 ~ 1,
  DM_INC == 2 ~ 2,
  DM_INC == 3 ~ 3,
  DM_INC == 4 ~ 4,
  DM_INC == 5 ~ 5,
  DM_INC == 6 ~ 6,
  DM_INC == 7 ~ 7,
  DM_INC == 8 ~ 8,
  DM_INC == 9 ~ 9,
  DM_INC == 10 ~ 10,
  DM_INC == 11 ~ 11,
  DM_INC == 12 ~ 12,
  DM_INC == 13 ~ 13,
  DM_INC == 14 ~ 14,
  DM_INC == 15 ~ 15,
  DM_INC == 16 ~ 16,
  DM_INC == 17 ~ 17,
  DM_INC == 18 ~ 18,
  DM_INC == 19 ~ 19,
  DM_INC == -7 ~ NA_real_,
  DM_INC == -4 ~ NA_real_)) %>% 
  mutate(View_of_the_Bible = if_else(ARBL_3 == 1, 1, 0)) %>% 
  select(Muslim_Hostility, Muslim_Contact, Punishing_God, Evangelical_Protestant, 
         Mainline_Protestant, Black_Protestant, Catholic, Jewish, Other_Religion, 
         Other_Protestant, Unaffiliated, Female, Age, Married, White, South, Republican, 
         Income, Education, View_of_the_Bible)

# Building summary table
table <- skim(PALS_2006)

summary_table <- table %>% mutate(N = nrow(PALS_2006) - n_missing) %>% 
  select(skim_variable, numeric.mean, numeric.sd, N) %>% 
  mutate(Mean = round(numeric.mean, 2)) %>% 
  mutate(Standard_Deviation = round(numeric.sd, 2)) %>% 
  select(skim_variable, Mean, Standard_Deviation, N)
```

```{r, warning=FALSE}
# Final summary table
summary_table %>% gt() %>% 
  tab_header(title = md("**Table 1**"), 
             subtitle = "Descriptive statistics of the dependent, independent, and control variables") %>% 
  tab_source_note(source_note = "Source: The Portraits of American Life Study 2006") %>% 
  tab_row_group(group = "Control Variables", rows = 12:20) %>% 
  tab_row_group(group = "Religious Affiliation", rows = 4:11) %>% 
  tab_row_group(group = "Independent Variables", rows = 2:3) %>% 
  tab_row_group(group = "Dependent Variable", rows = 1) %>% 
  cols_label(skim_variable = "")
```

&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;Following the essential summary statistics reported in Table 1, Jung estimates 5 different binomial logistic regressions to examine which factors have a significant effect on the odds of holding the least respect for Islam, with Models 2-5 meant to answer the hypotheses generated earlier. Model 1 simply regresses our set of controls for having the least respect for Islam. Model 2 adds our religious affiliation variables to the orginal set of controls in Model 1. Model 3 adds on to Model 2 by testing the significance of contact with Muslims on tolerance. Similarly to Model 3, Model 4 adds on to Model 2 by testing the signifance of one's image of God on tolerance. Lastly, Model 5 uses all the variables perviously listed adds on centered interaction terms between each relgious affiliation and level of Muslim contact. Model 5 investigates whether increased Muslim contact actually leads to more hostility among stricter religious groups such as Evangelicals. Jong centers these interaction terms in order to avoid multicollinearity Unfortunately, as Jung has not provided the R code used to generate his regressions and estimates, I have approximated his models and interactions through the listed variables on Table 2 in conjunction with the Gov 52 teaching team. Odds ratios above 1 indicate a greater likelihood for an individual to hold a low regard for Islam, while an odds ratio smaller than 1 indicates a decreased likelihood for discontent for Islam.

```{r}
# Model 1
Model_1 <- glm(Muslim_Hostility ~ Female + Age + Married + White + South + 
                 Republican + Income + Education + View_of_the_Bible, 
               data = PALS_2006, family = "binomial"(link='logit'))

# Model 2
Model_2 <- glm(Muslim_Hostility ~ Female + Age + Married + White + South + 
                 Republican + Income + Education + View_of_the_Bible + 
                 Evangelical_Protestant + Mainline_Protestant + 
                 Black_Protestant + Catholic + Jewish + Other_Religion + 
                 Other_Protestant + Unaffiliated, data = PALS_2006, 
               family = "binomial"(link='logit'))

# Model 3
Model_3 <- glm(Muslim_Hostility ~ Female + Age + Married + White + South + 
                 Republican + Income + Education + View_of_the_Bible + 
                 Evangelical_Protestant + Mainline_Protestant + 
                 Black_Protestant + Catholic + Jewish + Other_Religion + 
                 Other_Protestant + Unaffiliated + Muslim_Contact, 
               data = PALS_2006, family = "binomial"(link='logit'))

# Model 4
Model_4 <- glm(Muslim_Hostility ~ Female + Age + Married + White + South + 
                 Republican + Income + Education + View_of_the_Bible + 
                 Evangelical_Protestant + Mainline_Protestant + 
                 Black_Protestant + Catholic + Jewish + Other_Religion + 
                 Other_Protestant + Unaffiliated + Punishing_God, 
               data = PALS_2006 , family = "binomial"(link='logit'))

# Model 5

# Centering interaction terms
PALS_2006 <- PALS_2006 %>% 
  mutate(EP_Contact = Evangelical_Protestant * Muslim_Contact) %>% 
  mutate(Evangelical_Protestant_Muslim_Contact_Centered = EP_Contact - mean(EP_Contact, na.rm = TRUE)) %>% 
  mutate(MP_Contact = Mainline_Protestant * Muslim_Contact) %>% 
  mutate(Mainline_Protestant_Muslim_Contact_Centered = MP_Contact - mean(MP_Contact, na.rm = TRUE)) %>% 
  mutate(BP_Contact = Black_Protestant * Muslim_Contact) %>% 
  mutate(Black_Protestant_Muslim_Contact_Centered = BP_Contact - mean(BP_Contact, na.rm = TRUE)) %>% 
  mutate(Catholic_Contact = Catholic * Muslim_Contact) %>% 
  mutate(Catholic_Muslim_Contact_Centered = Catholic_Contact - mean(Catholic_Contact, na.rm = TRUE)) %>% 
  mutate(Jewish_Contact = Jewish * Muslim_Contact) %>% 
  mutate(Jewish_Muslim_Contact_Centered = Jewish_Contact - mean(Jewish_Contact, na.rm = TRUE)) %>% 
  mutate(OR_Contact = Other_Religion * Muslim_Contact) %>% 
  mutate(Other_Religion_Muslim_Contact_Centered = OR_Contact - mean(OR_Contact, na.rm = TRUE)) %>% 
  mutate(OP_Contact = Other_Protestant * Muslim_Contact) %>% 
  mutate(Other_Protestant_Muslim_Contact_Centered = OP_Contact - mean(OP_Contact, na.rm = TRUE)) %>% 
  mutate(U_Contact = Unaffiliated * Muslim_Contact) %>% 
  mutate(Unaffiliated_Muslim_Contact_Centered = U_Contact - mean(U_Contact, na.rm = TRUE))

Model_5 <- glm(Muslim_Hostility ~ Female + Age + Married + White + South + 
                 Republican + Income + Education + View_of_the_Bible + 
                 Evangelical_Protestant + Mainline_Protestant + Black_Protestant + 
                 Catholic + Jewish + Other_Religion + Other_Protestant + 
                 Unaffiliated + Muslim_Contact + Punishing_God + 
                 Evangelical_Protestant_Muslim_Contact_Centered + 
                 Mainline_Protestant_Muslim_Contact_Centered + 
                 Black_Protestant_Muslim_Contact_Centered + 
                 Catholic_Muslim_Contact_Centered + 
                 Jewish_Muslim_Contact_Centered + 
                 Other_Religion_Muslim_Contact_Centered + 
                 Other_Protestant_Muslim_Contact_Centered + 
                 Unaffiliated_Muslim_Contact_Centered, data = PALS_2006, 
               family = "binomial"(link='logit'))
```

```{r, results="asis", warning=FALSE}
# Creating combined table
stargazer2 <- function(model, odd.ratio = F, ...) {
  if(!("list" %in% class(model))) model <- list(model)
    
  if (odd.ratio) {
    coefOR2 <- lapply(model, function(x) exp(coef(x)))
    seOR2 <- lapply(model, function(x) exp(coef(x)) * summary(x)$coef[, 2])
    p2 <- lapply(model, function(x) summary(x)$coefficients[, 4])
    stargazer(model, coef = coefOR2, se = seOR2, p = p2, ...)
    
  } else {
    stargazer(model, ...)
  }
}

models <- list(Model_1, Model_2, Model_3, Model_4, Model_5)
stargazer2(models, title="Table 2: Regression Results", 
          align=TRUE, dep.var.labels= "Least Respect for Islam", 
          omit.stat=c("LL","ser","f"), no.space=TRUE, column.sep.width = "-15pt", 
          header = FALSE, font.size = "small", odd.ratio = T)
```

&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;Just as Jong Hyun Jung did in his 2012 paper, we find support for each of the hypotheses that he laid out. Religious affiliation indeed appears to be related to the degree to which one respects Islam, with Christian traditions including Evangelical, Mainline Protestant, and Catholic being significantly more likely to have the least respect for Islam (Hypothesis 1). I also find that the image of a punishing God is positively associated with the likelihood of Islam being least respected, although unlike Jung, I do not find that this significant at the 95% level (Hypothesis 2). Additionaly, greater contact with Muslims appears to significantly lower the likelihood of having the least respect for Islam (Hypothesis 3). I also find, like Jung, that Evangelical and Black Protestants display the opposite trend; increased contact with Muslims results in greater odds of having the least respect for Islam, although I do not find the Evangelical estimate significant as Jung did (Hypothesis 4). 

# An Extentsion of the Paper

&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;A general limitation I find in the paper lies in the time the data was collected. Measuring Islamophobia 14 years ago may not be generalizable today. The 2006 responses that Jong Hyun Jung utilizes were collected only five years after the attacks on 9/11 and the United States was still at war. While this gap may lessen the anti-Muslim biased sentiment immediately following the attacks, there is no telling how much of a confounding effect it may hold. Around the time Jung's paper was published, a follow-up PALS survey was released in 2012. This could possibly speak much clearer of the true causes of modern-day Islamophobia. Other than simply recalculating our regression estimates using the 2012 data, I propose a further Muslim contact analysis that I believe is worth examining. In response to rising Islamophobia of the early and mid-2000s, many Muslims around the United States have set up “Meet a Muslim!” campaigns, where communities get to sit down with local Muslims, ask questions, and strike up regular conversation. I've even had the priviledge of being part of these events while I was growing up. Many Muslims felt the need to engage with their fellow Americans to dispel negative perceptions of their faith, particularly as the country has seen a recent uptick in anti-Muslim crimes. The idea is that by myth-busting and getting to know people on a person-to-person basis it’s harder to demonize them. This begs the question, for the men, women, and children hoping to cast their religion in a better light, does slowly exposing individuals to more frequent Muslim contact over consecutive years work? 

&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;To estimate the causal impact of increased contact with Muslims over time, I will be using the merged PALS dataset which combines responses from the 2006 and 2012 surveys. For continuity, I'll be using a binary logistic regression with odds ratios and the same control variables Jung uses in his estimate of Muslim contact in Model 3. Below, you'll find the code I used to clean and recode the merged dataset, and create what I call 'Model 6', my logistic regression.

```{r}
# Reading in merged data
PALS_MERGED <- read_xlsx("Data/Portraits of American Life Study, Merged Dataset, 2006-2012.XLSX")

# Cleaning and recoding data
PALS_MERGED <- PALS_MERGED %>% select(RESP_ZRI, RC_2, IC_6W1, IC_6, 
                                      RELTRAD, RE_RACE, I_GENDER, EDUC, 
                                      I_PARTYID, REGION, DMINC, ARA5, 
                                      LV_1, ARBL3, HR_AGE) %>% 
  drop_na() %>% 
  mutate(contact_difference = IC_6 - IC_6W1) %>% 
  mutate(Contact_Increase = if_else(contact_difference > 0, 1, 0)) %>% 
  mutate(Muslim_Hostility = if_else(RC_2 == 4, 1, 0)) %>%
  mutate(Evangelical_Protestant = if_else(RELTRAD == 2, 1, 0)) %>% 
  mutate(Mainline_Protestant = if_else(RELTRAD == 3, 1, 0)) %>%
  mutate(Black_Protestant = if_else(RELTRAD == 1, 1, 0)) %>%
  mutate(Catholic = if_else(RELTRAD == 4, 1, 0)) %>%
  mutate(Jewish = if_else(RELTRAD == 5, 1, 0)) %>%
  mutate(Other_Religion = if_else(RELTRAD == 6, 1, 0)) %>%
  mutate(Other_Protestant = if_else(RELTRAD == 8, 1, 0)) %>%
  mutate(Unaffiliated = if_else(RELTRAD == 7, 1, 0)) %>%
  mutate(Punishing_God = case_when(
  ARA5 == 1 ~ 1,
  ARA5 == 2 ~ 2,
  ARA5 == 3 ~ 3,
  ARA5 == 4 ~ 4,
  ARA5 == 5 ~ 5,
  ARA5 == -7 ~ NA_real_,
  ARA5 == -4 ~ NA_real_)) %>% 
  mutate(White = if_else(RE_RACE == 1, 1, 0)) %>% 
  mutate(Female = if_else(I_GENDER == 0, 1, 0)) %>% 
  mutate(Education = case_when(
  EDUC == 1 ~ 1,
  EDUC == 2 ~ 2,
  EDUC == 3 ~ 3,
  EDUC == 4 ~ 4,
  EDUC == 5 ~ 5,
  EDUC == 6 ~ 6,
  EDUC == 7 ~ 7,
  EDUC == 8 ~ 8,
  EDUC == 9 ~ 9,
  EDUC == 10 ~ 10,
  EDUC == 11 ~ 11,
  EDUC == 12 ~ NA_real_)) %>% 
  mutate(Republican = if_else(I_PARTYID == 3, 1, 0)) %>% 
  mutate(South = if_else(REGION == 3, 1, 0)) %>% 
  mutate(Married = if_else(LV_1 == 1, 1, 0)) %>% 
  mutate(Income = case_when(
  DMINC == 1 ~ 1,
  DMINC == 2 ~ 2,
  DMINC == 3 ~ 3,
  DMINC == 4 ~ 4,
  DMINC == 5 ~ 5,
  DMINC == 6 ~ 6,
  DMINC == 7 ~ 7,
  DMINC == 8 ~ 8,
  DMINC == 9 ~ 9,
  DMINC == 10 ~ 10,
  DMINC == 11 ~ 11,
  DMINC == 12 ~ 12,
  DMINC == 13 ~ 13,
  DMINC == 14 ~ 14,
  DMINC == 15 ~ 15,
  DMINC == 16 ~ 16,
  DMINC == 17 ~ 17,
  DMINC == 18 ~ 18,
  DMINC == 19 ~ 19,
  DMINC == -7 ~ NA_real_,
  DMINC == -4 ~ NA_real_)) %>%
  mutate(Age = HR_AGE) %>% 
  mutate(View_of_the_Bible = if_else(ARBL3 == 1, 1, 0)) %>% 
  select(Muslim_Hostility, Contact_Increase, Punishing_God, Evangelical_Protestant, 
         Mainline_Protestant, Black_Protestant, Catholic, Jewish, Other_Religion, 
         Other_Protestant, Unaffiliated, Female, Age, Married, White, South, Republican, 
         Income, Education, View_of_the_Bible)
```

```{r, results='asis', warning=FALSE}
# Estimating Model 6
Model_6 <- glm(Muslim_Hostility ~ Female + Age + Married + White + South + 
                 Republican + Income + Education + View_of_the_Bible + 
                 Evangelical_Protestant + Mainline_Protestant + 
                 Black_Protestant + Catholic + Jewish + Other_Religion + 
                 Other_Protestant + Unaffiliated + Contact_Increase, 
               data = PALS_MERGED, family = "binomial"(link='logit'))

stargazer2(Model_6, title="Table 3: Model 6 Regression Results", 
          align=TRUE, dep.var.labels= "Least Respect for Islam", 
          omit.stat=c("LL","ser","f"), no.space=TRUE, column.sep.width = "-15pt", 
          header = FALSE, font.size = "small", odd.ratio = T)
```

&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;In Model 6, I’ve added a variable measuring if an individual has increased contact in everyday life with Muslims between 2006 and 2012 (Contact_Increase). As we can see, increased contact with Muslims over that six year period is associated with decreased odds of hostility towards Islam. However, we also find that this reduction is not significant at the 95% level. I found this surprising as our previous models suggested that contact of any kind was highly significant in reducing hostility. I was always taught that increased contact with Muslims over time can lead to more information and enhanced intercultural understanding about us, which can reduce bias by increasing recognition and succesfully counter Islamophobia. It appears that simply increasing contact may not be the only key to the puzzle, and perhaps countering Islamophobia requires highly regular contact in general between Muslims and non-Muslims to effectively dispel myths and stereotypes.

# Works Cited

Bakali, N. (2016). Islamophobia : Understanding Anti-Muslim Racism through the Lived Experiences of Muslim Youth (Transgressions: Cultural Studies and Education; v. 116). Leiden, Boston: Brill | Sense.

Bubenik, T. (2017, March 11). Youth Hold 'Meet A Muslim' Events To Encourage People To Learn About Each Other. Retrieved from https://www.npr.org/2017/03/11/519845905/youth-hold-meet-a-muslim-events-to-encourage-people-to-learn-about-each-other

Campbell, D. (2006). Religious “Threat” in Contemporary Presidential Elections. The Journal of Politics, 68(1), 104-115.

Cimino, R. (2005). "No God in Common:" American Evangelical Discourse on Islam after 9/11. Review of Religious Research, 47(2), 162-174.

Edgell, P., Gerteis, J., & Hartmann, D. (2006). Atheists As "Other": Moral Boundaries and Cultural Membership in American Society. American Sociological Review, 71(2), 211-234.

Emerson, Michael O., and David Sikkink. Portraits of American Life Study, 1st Wave, 2006.

Finke, R., & Stark, R. (2005). The Churching of America, 1776-2005 :Winners and Losers in Our Religious Economy (2nd ed.). New Brunswick, N.J.: Rutgers University Press.

Khari Brown, R., & Brown, Ronald. (2011). The Challenge of Religious Pluralism: The Association Between Interfaith Contact and Religious Pluralism. Review of Religious Research, 53(3), 323-340.

Marty, M. (2004). The Protestant Voice in American Pluralism. University of Georgia Press.

Mohamed, B. (2018, January 3). A new estimate of U.S. Muslim population. Retrieved from https://www.pewresearch.org/fact-tank/2018/01/03/new-estimates-show-u-s-muslim-population-continues-to-grow/

Pew Research. (2010). NYC Mosque Opposed, Muslims’ Right to Build Mosques Favored: Public Remains Conflicted Over Islam. Retrieved from https://www.pewresearch.org/wp-content/uploads/sites/7/2010/08/Islam-mosque-full-report.pdf

Putnam, R., & Campbell, D. (2012). American Grace: How religion divides and unites us / (1. Simon & Schuster trade pbk. ed.). New York: Simon & Schuster.

Wald, K., & Calhoun-Brown, A. (2011). Religion and politics in the United States (6th ed.). Lanham, Md.: Rowman & Littlefield.


[^1]: All data and code used in this replication project can be obtained 
[^2]: Jung, Jong Hyun. 2012. "Islamophobia?: Religion, Contact with Muslims, and the Respect for Islam." Review of Religious Research, 54(1): 113-126.



















